Instant Casting Movie Theater: The Future Cast System

Facial animation reflecting personal characteristics by automatic head modeling and facial muscle adjustment

“…In the earlier phase of our system development, using the low-cost range scanner, we needed to generate a subject's CG character that they can identify themselves in the film accurately. The experimental result shows that the FCS succeeds to generate an face model at 95% for 1630,000 audiences regardless of a male/female and young/elderly, in fact, however, 40% of audiences couldn't identify themselves in the film [7]. This is because the generated CG character is lack of subjects' "Personal Characteristics" such as body shape, walking style, head shape, hairstyle, expression style, voice characteristic, accessories, and so on.…”

“…An acquired geometry usually has large holes in the hair region. For these reasons, we use the facial range scanner [7] to acquire a facial geometry, and complement the geometry where is impossible to capture using a template mesh geometry.…”

“…An individual facial muscle layout is determined by categorizing the pattern of facial movement obtained from a subject's facial expression videos, selecting appropriate one from variation of pre-designed facial muscle layouts. The facial animation can be created by blending individual key shapes according to blend weights which are defined along with scenario in the same manner [7]. As a result, we can synthesize a CG character with a person-specific facial expression.…”

“…These methods require manual tunings such as the model/camera pose, orientation and the camera focal length parameter settings or marker specification. Unlike these approaches, our method can full-automatically create an individual head model based on the automatic facial feature point detector [7].…”

A SfM-based 3D face reconstruction method robust to self-occlusion by using a shape conversion matrix

“…Modeling 3D faces is a very useful technology in computer vision for its various applications, including pose-invariant face recognition [1,2], age-invariant face recognition [3], 3D personspecific game and movie character generation [4,5], teleconferencing, surgical simulation, etc. One approach used to model a 3D face requires special hardware or multi-cameras, such as a 3D laser scanner, stereo cameras, or a structured light [6].…”